After nearly 8 years of blogging, my post categories have turned out to be rather futile. If I wanted to curate a coherent assembly, I would need new categories or tags or keywords. But soon I was reminded again of Stephen Downes’ explanation: “the problem is that you’re tied to the sign, word and symbol.” (which bothered me before). I ended up with unnamed clusters of tags — because the interesting concepts lie between these tags, in the connections.
The first problem was that blog posts are much different than articles: Blog posts are tightly tied into contemporary external contexts and densely connected to each other, which means that they are easily accessible within that context, but much harder as time goes by. Other articles are more independent of time and context, reusable and “sterile” [Downes 2012], p. 9), because they are wrapped and armored with plenty of introductory and concluding statements. So to speak, they carry their “root ball” with themselves like a plant from the tree nursery which is going to be repotted. By contrast, blog posts are more comparable to rhizomes that cannot easily be pulled out without their neighboring ones.
So it took me many attempts over many weeks, to find a way how to select connections that would ideally allow for a coherent narrative.
Taking the tags that connected two linked blog posts, led me to the futile stage 2 where even tools like NetDraw did not help to untangle the felted mesh. Next I tried to isolate a chain of linked tags, see what looks like a ring road in stage 3. In stage 4 I sacrificed many connections and put up with the idea of a hierarchical ranking of tags and backbone of links. But I still could not find labels for the emerging topics. (I learned some hidden benefits of Cmap: After exporting the links between my posts, “propositions as text”, I could transfer them via Excel to Access, count them, and transfer them via database queries and proposition import to a smaller Cmap again — stage 5). Finally, I put boxes (Cmap “containers”) around my unnamed clusters, and let go the idea of short label words, and contented myself with header lines that mention multiple tags. This made my database manipulations for producing the HTML output more difficult, but such a little technical problem is more welcome than the big conceptual problem mentioned above.